基于纹理与颜色特征融合的刑侦图像检索算法  

A retrieval algorithm of forensic image fusion based on fusioning texture and color feature

作　　者：兰蓉[1] 贾世英 

机构地区：[1]西安邮电大学通信与信息工程学院,陕西西安710121

出　　处：《西安邮电大学学报》2016年第2期57-62,共6页Journal of Xi’an University of Posts and Telecommunications

基　　金：陕西省教育厅科学研究计划资助项目(15JK1658)

摘　　要：针对刑侦图像检索,给出一种基于多方向多参数的灰度共生矩阵和HSV颜色矩的图像特征提取算法。计算刑侦图像8个方向上6个参数,共48个特征值来表示图像的纹理特征,分别计算图像H、S和V通道的一阶矩、二阶矩和三阶矩,作为图像的颜色特征,并将纹理特征和颜色特征进行融合以保证特征的全面性。对刑侦图像数据库进行检索试验,结果表明,相较于4参数的灰度共生矩阵法和小波变换融合颜色直方图检索方法,所给方法的各类图像查准率更高。以指纹类别为例,所给算法的查准率为68%,高于4参数灰度共生矩阵算法的49%,和小波变换融合颜色直方图检索算法的53%。For criminal investigation image retrieval,an image feature extraction algorithm based on multi direction and multi parameters gray level co-occurrence matrix and HSV color moment is proposed.6parameters in each of 8directions on a criminal investigation image are calculated,thus,48 feature values can be used to represent the texture feature of the image.The first,second and third moment of H,S,and V channel of the image are calculated,thus,9feature values can be used to discribe the color feature of the image.All these texture features and color features are fused together,thus,the image can be discribed more comprehensive.A retrieval test is done on the criminal investigation image database,and the results show that,compared with the four parameters gray level co-occurrence matrix algorithm and wavelet transform fusion color histogram retrieval algorithm,the proposed algorithm has the highest precision,as for fingerprint images,the precision by the improved algorithm is 68%,higher than 49% and 53%respectively by the other two algorithms.

关 键 词：灰度共生矩阵 颜色特征 图像检索 相似性度量 

分 类 号：TN911.7[电子电信—通信与信息系统]